As information and communication networks are highly interconnected with the power grid, cyber security of the supervisory control and data acquisition (SCADA) system, or energy management system (EMS), has become a critical issue in the electric power sector. Cyber security of the SCADA/EMS system can be improved by deploying the trust systems to monitor network traffic for malicious attacks. However, cyber attacks could still be successful if attackers possess sufficient offensive resources. By exploiting vulnerabilities of cyber components, attackers are able to intrude into the SCADA/EMS system and send fabricated commands to local field devices. The reliability of the power system could thus be impacted by various cyber attacks. Quantitative assessment of the cyber attack and defense on the SCADA/EMS system are needed in cyber security evaluation approaches. For instance, the deployment and runtime cost due to the defensive trust systems will affect the original function of the power system without careful planning and design. Therefore, optimization of static trust system placement and routes between nodes, as well as self-recovery mechanisms, should be considered in the SCADA/ EMS network. Furthermore, vulnerability evaluation of the cyber attacks on the power system should be analyzed. A number of methods are used to characterize the attack patterns, and estimate frequencies of the cyber attack that may occur on the SCADA/EMS system. Examples include Common Vulnerability Scoring System (CVSS), Attack Graph, and Mean Time-to-Compromise (MTTC). Meanwhile, probabilistic methods used for reliability evaluation, such as Monte Carlo simulation (MCS), should be improved by considering impacts of cyber attacks on the power system. This dissertation is focused on optimizing both the static placement of the trust nodes and the dynamic routing of communicating nodes in the SCADA/EMS system, as well as investigating the reliability impact of the power system considering probabilities of successful cyber attacks on the SCADA/EMS system. Contributions of this dissertation include optimization of the trust system placement in the SCADA/EMS system using the set packing algorithm, the least cost route searching with the trust node aware optimal routing algorithm, the use of Bayesian attack graph and MTTC models to demonstrate probabilities of successful cyber attacks, modeling of normal and penetration attacks against the intrusion tolerant system with a modified semi-Markov process (SMP) model, estimation of MTTCs on the control center and substations considering the optimal allocation of offensive and defensive resources, power system reliability analysis incorporating cyber attacks in the integrated wind farm SCADA/EMS system architecture, and estimation of reliability indices in IEEE reliability test systems considering cyber attacks on SCADA/EMS system.